Radiated Emc ethernet problems

I have just designed a prototype Arm 9 based board that uses a Davicom DM9161 ethernet chip operating in RM11 mode, basically it require a external 50MHz oscillator that feeds it and the MAC in the processor. The board has just failed EMC emissions at 50MHz and 100MHz which are due to the ethernet, disconnectioning but still running the 50MHz ethernet oscillator made it pass. We have the option on the redesign of using MII ethernet mode, which uses a 25Mhz crystal into the DM9161, though internally a 100MHz clock is still required, the communications to the MAC is then at 25MHz (100Base) and comes from the DM9161 TXCLK and RXCLK. Would you expect doing this would reduce the EMC emission problems at 50 and 100MHz? Thanks for any advice

Steve jones

Reply to
sjones
Loading thread data ...

First thing is to check your PCB layout. Are you following all the rules? The layout on the other side of DM9161 is even more important, i.e. the magnetic, RJ-45, etc. Can you post your layout somewhere? Or by email?

Reply to
linnix

Yes, have followed all the recommended layout rules. 6 Layer PCB. Analogue planes under chip, fed via ferrite beads from 3V3 and 0V. Lots of decoupling. RJ45 with integrated magnetics.. Chip within 20mm of magnetics. 50MHz oscillator next to chip. MAC about 75mm away.

Reply to
sjones

Could email you directly. What is your experience with this chip and emc?

Reply to
sjones

Have done some designs with this and similar chip. It does not like ummatched distances (too far or too close), signal cross-over and others. How about RF shielding over and under the PCB near the chip? Are you using a metal enclosure? You can email me (using my initials) at linnix dot com

Thanks.

epl

Reply to
linnix

We got a design using the DM9161 to pass CE and FCC class b with an unshielded ethernet cable in MII mode, but the 125Mhz Ethernet frame rate was damm close.

Make sure the thing you are connecting to externally is VERY clean. Ended up connecting it to a battery powered laptop in a Farafay cage.

The Test house (TUV) said we could do anything we wanted in the cage, but had to have at least one meter of cable exposed between the farady cage and the DUT. We tested several laptops, choose the quietest one and put commmon mode filters on the cable inside the cage.

Not sure it's completly kosher, but we passed.

We also tested a competitors product under the same conditions and even though it had a CE mark it was out of spec by 20dB or more.

Go figure?

Paul

P.S. I'd rather be lucky than good. In the screen room pretest we had one birdy that was possibly close. When we go to the open field site, there was a 50KW FM station right exactly on that frequecy the frequency was not one that made sense to me so maybe the FM station was leaking into the screen room Duh just though of that ;) .

Reply to
Paul

Thanks, did you have the product in a plastic or metal case?

Reply to
sjones

We have gotten multiple things to pass with this setup a couple of examples:.

See:

formatting link
(Metal case)

and

Open board on the open carrier board with no case at all.

formatting link

Paul (Rasdoc is my home account)

Reply to
pbreed

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.